ComfyUI Node
Krea 2 Moodboard
A ComfyUI node in conditioning/krea2 with 11 inputs and 1 output.
Krea 2 Moodboard
- clip
- image
- image2
- CONDITIONING
◄prompt—►
◄strength0.50►
◄extract▾►
◄position▾►
◄reference_processing▾►
◄indirectfalse►
◄style_directivetrue►
◄vision_px1024►
Categoryconditioning/krea2
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| clip | CLIP | — | |
| prompt | STRING | — | |
| strength | FLOAT | 0.500–1 | 1.0 = full reference detail (layout/pose can leak); lower = keep palette/texture/mood only. |
| imageopt | IMAGE | reference image; leave unconnected for a plain Krea 2 text encode | |
| image2opt | IMAGE | — | |
| extractopt | COMBO | style = palette/lighting/texture; subject = composition/content, prompt controls the look. | |
| positionopt | COMBO | Vision tokens before or after the prompt. 'after' keeps the prompt image-blind (less content leak). | |
| reference_processingopt | COMBO | Shuffled crops kill composition, keep palette/texture. Crops make several vision spans: if outputs turn into grids, enable indirect. | |
| indirectopt | BOOLEAN | false | Delete vision rows after encoding: style only via prompt re-contextualization. Strongest anti-leak mode; forces 'before' position. |
| style_directiveopt | BOOLEAN | true | Inject a declarative style-not-content (or subject) sentence next to the vision span. |
| vision_pxopt | INT | 10240–4096 | Cap on the longest side fed to Qwen3-VL. 0 = never resize. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| CONDITIONING | CONDITIONING | — |